5 * This file has definitions for some important file table
9 #include <linux/limits.h>
10 #include <linux/ioctl.h>
11 #include <linux/blk_types.h>
14 * It's silly to have NR_OPEN bigger than NR_FILE, but you can change
15 * the file limit at runtime and only root can increase the per-process
16 * nr_file rlimit, so it's safe to set up a ridiculously high absolute
17 * upper limit on files-per-process.
19 * Some programs (notably those using select()) may have to be
20 * recompiled to take full advantage of the new limits..
23 /* Fixed constants first: */
25 #define INR_OPEN 1024 /* Initial setting for nfile rlimits */
27 #define BLOCK_SIZE_BITS 10
28 #define BLOCK_SIZE (1<<BLOCK_SIZE_BITS)
30 #define SEEK_SET 0 /* seek relative to beginning of file */
31 #define SEEK_CUR 1 /* seek relative to current file position */
32 #define SEEK_END 2 /* seek relative to end of file */
33 #define SEEK_MAX SEEK_END
35 /* And dynamically-tunable limits and defaults: */
36 struct files_stat_struct {
37 int nr_files; /* read only */
38 int nr_free_files; /* read only */
39 int max_files; /* tunable */
42 struct inodes_stat_t {
45 int dummy[5]; /* padding for sysctl ABI compatibility */
49 #define NR_FILE 8192 /* this can well be larger on a larger system */
60 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
61 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
64 /* file is open for reading */
65 #define FMODE_READ ((__force fmode_t)0x1)
66 /* file is open for writing */
67 #define FMODE_WRITE ((__force fmode_t)0x2)
68 /* file is seekable */
69 #define FMODE_LSEEK ((__force fmode_t)0x4)
70 /* file can be accessed using pread */
71 #define FMODE_PREAD ((__force fmode_t)0x8)
72 /* file can be accessed using pwrite */
73 #define FMODE_PWRITE ((__force fmode_t)0x10)
74 /* File is opened for execution with sys_execve / sys_uselib */
75 #define FMODE_EXEC ((__force fmode_t)0x20)
76 /* File is opened with O_NDELAY (only set for block devices) */
77 #define FMODE_NDELAY ((__force fmode_t)0x40)
78 /* File is opened with O_EXCL (only set for block devices) */
79 #define FMODE_EXCL ((__force fmode_t)0x80)
80 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
81 (specialy hack for floppy.c) */
82 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
85 * Don't update ctime and mtime.
87 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
88 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
90 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
92 /* Expect random access pattern */
93 #define FMODE_RANDOM ((__force fmode_t)0x1000)
95 /* File was opened by fanotify and shouldn't generate fanotify events */
96 #define FMODE_NONOTIFY ((__force fmode_t)0x1000000)
99 * The below are the various read and write types that we support. Some of
100 * them include behavioral modifiers that send information down to the
101 * block layer and IO scheduler. Terminology:
103 * The block layer uses device plugging to defer IO a little bit, in
104 * the hope that we will see more IO very shortly. This increases
105 * coalescing of adjacent IO and thus reduces the number of IOs we
106 * have to send to the device. It also allows for better queuing,
107 * if the IO isn't mergeable. If the caller is going to be waiting
108 * for the IO, then he must ensure that the device is unplugged so
109 * that the IO is dispatched to the driver.
111 * All IO is handled async in Linux. This is fine for background
112 * writes, but for reads or writes that someone waits for completion
113 * on, we want to notify the block layer and IO scheduler so that they
114 * know about it. That allows them to make better scheduling
115 * decisions. So when the below references 'sync' and 'async', it
116 * is referencing this priority hint.
118 * With that in mind, the available types are:
120 * READ A normal read operation. Device will be plugged.
121 * READ_SYNC A synchronous read. Device is not plugged, caller can
122 * immediately wait on this read without caring about
124 * READA Used for read-ahead operations. Lower priority, and the
125 * block layer could (in theory) choose to ignore this
126 * request if it runs into resource problems.
127 * WRITE A normal async write. Device will be plugged.
128 * WRITE_SYNC_PLUG Synchronous write. Identical to WRITE, but passes down
129 * the hint that someone will be waiting on this IO
130 * shortly. The device must still be unplugged explicitly,
131 * WRITE_SYNC_PLUG does not do this as we could be
132 * submitting more writes before we actually wait on any
134 * WRITE_SYNC Like WRITE_SYNC_PLUG, but also unplugs the device
135 * immediately after submission. The write equivalent
137 * WRITE_ODIRECT_PLUG Special case write for O_DIRECT only.
138 * WRITE_BARRIER Like WRITE_SYNC, but tells the block layer that all
139 * previously submitted writes must be safely on storage
140 * before this one is started. Also guarantees that when
141 * this write is complete, it itself is also safely on
142 * storage. Prevents reordering of writes on both sides
146 #define RW_MASK REQ_WRITE
147 #define RWA_MASK REQ_RAHEAD
150 #define WRITE RW_MASK
151 #define READA RWA_MASK
153 #define READ_SYNC (READ | REQ_SYNC | REQ_UNPLUG)
154 #define READ_META (READ | REQ_META)
155 #define WRITE_SYNC_PLUG (WRITE | REQ_SYNC | REQ_NOIDLE)
156 #define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_UNPLUG)
157 #define WRITE_ODIRECT_PLUG (WRITE | REQ_SYNC)
158 #define WRITE_META (WRITE | REQ_META)
159 #define WRITE_BARRIER (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_UNPLUG | \
163 * These aren't really reads or writes, they pass down information about
164 * parts of device that are now unused by the file system.
166 #define DISCARD_NOBARRIER (WRITE | REQ_DISCARD)
167 #define DISCARD_BARRIER (WRITE | REQ_DISCARD | REQ_HARDBARRIER)
168 #define DISCARD_SECURE (DISCARD_NOBARRIER | REQ_SECURE)
174 /* public flags for file_system_type */
175 #define FS_REQUIRES_DEV 1
176 #define FS_BINARY_MOUNTDATA 2
177 #define FS_HAS_SUBTYPE 4
178 #define FS_REVAL_DOT 16384 /* Check the paths ".", ".." for staleness */
179 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move()
180 * during rename() internally.
184 * These are the fs-independent mount-flags: up to 32 flags are supported
186 #define MS_RDONLY 1 /* Mount read-only */
187 #define MS_NOSUID 2 /* Ignore suid and sgid bits */
188 #define MS_NODEV 4 /* Disallow access to device special files */
189 #define MS_NOEXEC 8 /* Disallow program execution */
190 #define MS_SYNCHRONOUS 16 /* Writes are synced at once */
191 #define MS_REMOUNT 32 /* Alter flags of a mounted FS */
192 #define MS_MANDLOCK 64 /* Allow mandatory locks on an FS */
193 #define MS_DIRSYNC 128 /* Directory modifications are synchronous */
194 #define MS_NOATIME 1024 /* Do not update access times. */
195 #define MS_NODIRATIME 2048 /* Do not update directory access times */
199 #define MS_VERBOSE 32768 /* War is peace. Verbosity is silence.
200 MS_VERBOSE is deprecated. */
201 #define MS_SILENT 32768
202 #define MS_POSIXACL (1<<16) /* VFS does not apply the umask */
203 #define MS_UNBINDABLE (1<<17) /* change to unbindable */
204 #define MS_PRIVATE (1<<18) /* change to private */
205 #define MS_SLAVE (1<<19) /* change to slave */
206 #define MS_SHARED (1<<20) /* change to shared */
207 #define MS_RELATIME (1<<21) /* Update atime relative to mtime/ctime. */
208 #define MS_KERNMOUNT (1<<22) /* this is a kern_mount call */
209 #define MS_I_VERSION (1<<23) /* Update inode I_version field */
210 #define MS_STRICTATIME (1<<24) /* Always perform atime updates */
211 #define MS_BORN (1<<29)
212 #define MS_ACTIVE (1<<30)
213 #define MS_NOUSER (1<<31)
216 * Superblock flags that can be altered by MS_REMOUNT
218 #define MS_RMT_MASK (MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK|MS_I_VERSION)
221 * Old magic mount flag and mask
223 #define MS_MGC_VAL 0xC0ED0000
224 #define MS_MGC_MSK 0xffff0000
226 /* Inode flags - they have nothing to superblock flags now */
228 #define S_SYNC 1 /* Writes are synced at once */
229 #define S_NOATIME 2 /* Do not update access times */
230 #define S_APPEND 4 /* Append-only file */
231 #define S_IMMUTABLE 8 /* Immutable file */
232 #define S_DEAD 16 /* removed, but still open directory */
233 #define S_NOQUOTA 32 /* Inode is not counted to quota */
234 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
235 #define S_NOCMTIME 128 /* Do not update file c/mtime */
236 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
237 #define S_PRIVATE 512 /* Inode is fs-internal */
240 * Note that nosuid etc flags are inode-specific: setting some file-system
241 * flags just means all the inodes inherit those flags by default. It might be
242 * possible to override it selectively if you really wanted to with some
243 * ioctl() that is not currently implemented.
245 * Exception: MS_RDONLY is always applied to the entire file system.
247 * Unfortunately, it is possible to change a filesystems flags with it mounted
248 * with files in use. This means that all of the inodes will not have their
249 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
250 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
252 #define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg))
254 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
255 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
256 ((inode)->i_flags & S_SYNC))
257 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
258 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
259 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
260 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
261 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
263 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
264 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
265 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
266 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
268 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
269 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
270 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
271 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
273 /* the read-only stuff doesn't really belong here, but any other place is
274 probably as bad and I don't want to create yet another include file. */
276 #define BLKROSET _IO(0x12,93) /* set device read-only (0 = read-write) */
277 #define BLKROGET _IO(0x12,94) /* get read-only status (0 = read_write) */
278 #define BLKRRPART _IO(0x12,95) /* re-read partition table */
279 #define BLKGETSIZE _IO(0x12,96) /* return device size /512 (long *arg) */
280 #define BLKFLSBUF _IO(0x12,97) /* flush buffer cache */
281 #define BLKRASET _IO(0x12,98) /* set read ahead for block device */
282 #define BLKRAGET _IO(0x12,99) /* get current read ahead setting */
283 #define BLKFRASET _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */
284 #define BLKFRAGET _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */
285 #define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */
286 #define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */
287 #define BLKSSZGET _IO(0x12,104)/* get block device sector size */
289 #define BLKPG _IO(0x12,105)/* See blkpg.h */
291 /* Some people are morons. Do not use sizeof! */
293 #define BLKELVGET _IOR(0x12,106,size_t)/* elevator get */
294 #define BLKELVSET _IOW(0x12,107,size_t)/* elevator set */
295 /* This was here just to show that the number is taken -
296 probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */
298 /* A jump here: 108-111 have been used for various private purposes. */
299 #define BLKBSZGET _IOR(0x12,112,size_t)
300 #define BLKBSZSET _IOW(0x12,113,size_t)
301 #define BLKGETSIZE64 _IOR(0x12,114,size_t) /* return device size in bytes (u64 *arg) */
302 #define BLKTRACESETUP _IOWR(0x12,115,struct blk_user_trace_setup)
303 #define BLKTRACESTART _IO(0x12,116)
304 #define BLKTRACESTOP _IO(0x12,117)
305 #define BLKTRACETEARDOWN _IO(0x12,118)
306 #define BLKDISCARD _IO(0x12,119)
307 #define BLKIOMIN _IO(0x12,120)
308 #define BLKIOOPT _IO(0x12,121)
309 #define BLKALIGNOFF _IO(0x12,122)
310 #define BLKPBSZGET _IO(0x12,123)
311 #define BLKDISCARDZEROES _IO(0x12,124)
312 #define BLKSECDISCARD _IO(0x12,125)
314 #define BMAP_IOCTL 1 /* obsolete - kept for compatibility */
315 #define FIBMAP _IO(0x00,1) /* bmap access */
316 #define FIGETBSZ _IO(0x00,2) /* get the block size used for bmap */
317 #define FIFREEZE _IOWR('X', 119, int) /* Freeze */
318 #define FITHAW _IOWR('X', 120, int) /* Thaw */
320 #define FS_IOC_GETFLAGS _IOR('f', 1, long)
321 #define FS_IOC_SETFLAGS _IOW('f', 2, long)
322 #define FS_IOC_GETVERSION _IOR('v', 1, long)
323 #define FS_IOC_SETVERSION _IOW('v', 2, long)
324 #define FS_IOC_FIEMAP _IOWR('f', 11, struct fiemap)
325 #define FS_IOC32_GETFLAGS _IOR('f', 1, int)
326 #define FS_IOC32_SETFLAGS _IOW('f', 2, int)
327 #define FS_IOC32_GETVERSION _IOR('v', 1, int)
328 #define FS_IOC32_SETVERSION _IOW('v', 2, int)
331 * Inode flags (FS_IOC_GETFLAGS / FS_IOC_SETFLAGS)
333 #define FS_SECRM_FL 0x00000001 /* Secure deletion */
334 #define FS_UNRM_FL 0x00000002 /* Undelete */
335 #define FS_COMPR_FL 0x00000004 /* Compress file */
336 #define FS_SYNC_FL 0x00000008 /* Synchronous updates */
337 #define FS_IMMUTABLE_FL 0x00000010 /* Immutable file */
338 #define FS_APPEND_FL 0x00000020 /* writes to file may only append */
339 #define FS_NODUMP_FL 0x00000040 /* do not dump file */
340 #define FS_NOATIME_FL 0x00000080 /* do not update atime */
341 /* Reserved for compression usage... */
342 #define FS_DIRTY_FL 0x00000100
343 #define FS_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
344 #define FS_NOCOMP_FL 0x00000400 /* Don't compress */
345 #define FS_ECOMPR_FL 0x00000800 /* Compression error */
346 /* End compression flags --- maybe not all used */
347 #define FS_BTREE_FL 0x00001000 /* btree format dir */
348 #define FS_INDEX_FL 0x00001000 /* hash-indexed directory */
349 #define FS_IMAGIC_FL 0x00002000 /* AFS directory */
350 #define FS_JOURNAL_DATA_FL 0x00004000 /* Reserved for ext3 */
351 #define FS_NOTAIL_FL 0x00008000 /* file tail should not be merged */
352 #define FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
353 #define FS_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
354 #define FS_EXTENT_FL 0x00080000 /* Extents */
355 #define FS_DIRECTIO_FL 0x00100000 /* Use direct i/o */
356 #define FS_RESERVED_FL 0x80000000 /* reserved for ext2 lib */
358 #define FS_FL_USER_VISIBLE 0x0003DFFF /* User visible flags */
359 #define FS_FL_USER_MODIFIABLE 0x000380FF /* User modifiable flags */
362 #define SYNC_FILE_RANGE_WAIT_BEFORE 1
363 #define SYNC_FILE_RANGE_WRITE 2
364 #define SYNC_FILE_RANGE_WAIT_AFTER 4
368 #include <linux/linkage.h>
369 #include <linux/wait.h>
370 #include <linux/types.h>
371 #include <linux/kdev_t.h>
372 #include <linux/dcache.h>
373 #include <linux/path.h>
374 #include <linux/stat.h>
375 #include <linux/cache.h>
376 #include <linux/kobject.h>
377 #include <linux/list.h>
378 #include <linux/radix-tree.h>
379 #include <linux/prio_tree.h>
380 #include <linux/init.h>
381 #include <linux/pid.h>
382 #include <linux/mutex.h>
383 #include <linux/capability.h>
384 #include <linux/semaphore.h>
385 #include <linux/fiemap.h>
387 #include <asm/atomic.h>
388 #include <asm/byteorder.h>
390 struct export_operations;
395 struct pipe_inode_info;
396 struct poll_table_struct;
398 struct vm_area_struct;
402 extern void __init inode_init(void);
403 extern void __init inode_init_early(void);
404 extern void __init files_init(unsigned long);
406 extern struct files_stat_struct files_stat;
407 extern int get_max_files(void);
408 extern int sysctl_nr_open;
409 extern struct inodes_stat_t inodes_stat;
410 extern int leases_enable, lease_break_time;
413 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
414 struct buffer_head *bh_result, int create);
415 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
416 ssize_t bytes, void *private, int ret,
420 * Attribute flags. These should be or-ed together to figure out what
423 #define ATTR_MODE (1 << 0)
424 #define ATTR_UID (1 << 1)
425 #define ATTR_GID (1 << 2)
426 #define ATTR_SIZE (1 << 3)
427 #define ATTR_ATIME (1 << 4)
428 #define ATTR_MTIME (1 << 5)
429 #define ATTR_CTIME (1 << 6)
430 #define ATTR_ATIME_SET (1 << 7)
431 #define ATTR_MTIME_SET (1 << 8)
432 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
433 #define ATTR_ATTR_FLAG (1 << 10)
434 #define ATTR_KILL_SUID (1 << 11)
435 #define ATTR_KILL_SGID (1 << 12)
436 #define ATTR_FILE (1 << 13)
437 #define ATTR_KILL_PRIV (1 << 14)
438 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
439 #define ATTR_TIMES_SET (1 << 16)
442 * This is the Inode Attributes structure, used for notify_change(). It
443 * uses the above definitions as flags, to know which values have changed.
444 * Also, in this manner, a Filesystem can look at only the values it cares
445 * about. Basically, these are the attributes that the VFS layer can
446 * request to change from the FS layer.
448 * Derek Atkins <warlord@MIT.EDU> 94-10-20
451 unsigned int ia_valid;
456 struct timespec ia_atime;
457 struct timespec ia_mtime;
458 struct timespec ia_ctime;
461 * Not an attribute, but an auxilary info for filesystems wanting to
462 * implement an ftruncate() like method. NOTE: filesystem should
463 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
465 struct file *ia_file;
469 * Includes for diskquotas.
471 #include <linux/quota.h>
474 * enum positive_aop_returns - aop return codes with specific semantics
476 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
477 * completed, that the page is still locked, and
478 * should be considered active. The VM uses this hint
479 * to return the page to the active list -- it won't
480 * be a candidate for writeback again in the near
481 * future. Other callers must be careful to unlock
482 * the page if they get this return. Returned by
485 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
486 * unlocked it and the page might have been truncated.
487 * The caller should back up to acquiring a new page and
488 * trying again. The aop will be taking reasonable
489 * precautions not to livelock. If the caller held a page
490 * reference, it should drop it before retrying. Returned
493 * address_space_operation functions return these large constants to indicate
494 * special semantics to the caller. These are much larger than the bytes in a
495 * page to allow for functions that return the number of bytes operated on in a
499 enum positive_aop_returns {
500 AOP_WRITEPAGE_ACTIVATE = 0x80000,
501 AOP_TRUNCATED_PAGE = 0x80001,
504 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
505 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
506 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
507 * helper code (eg buffer layer)
508 * to clear GFP_FS from alloc */
511 * oh the beauties of C type declarations.
514 struct address_space;
515 struct writeback_control;
518 const struct iovec *iov;
519 unsigned long nr_segs;
524 size_t iov_iter_copy_from_user_atomic(struct page *page,
525 struct iov_iter *i, unsigned long offset, size_t bytes);
526 size_t iov_iter_copy_from_user(struct page *page,
527 struct iov_iter *i, unsigned long offset, size_t bytes);
528 void iov_iter_advance(struct iov_iter *i, size_t bytes);
529 int iov_iter_fault_in_readable(struct iov_iter *i, size_t bytes);
530 size_t iov_iter_single_seg_count(struct iov_iter *i);
532 static inline void iov_iter_init(struct iov_iter *i,
533 const struct iovec *iov, unsigned long nr_segs,
534 size_t count, size_t written)
537 i->nr_segs = nr_segs;
539 i->count = count + written;
541 iov_iter_advance(i, written);
544 static inline size_t iov_iter_count(struct iov_iter *i)
550 * "descriptor" for what we're up to with a read.
551 * This allows us to use the same read code yet
552 * have multiple different users of the data that
553 * we read from a file.
555 * The simplest case just copies the data to user
568 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
569 unsigned long, unsigned long);
571 struct address_space_operations {
572 int (*writepage)(struct page *page, struct writeback_control *wbc);
573 int (*readpage)(struct file *, struct page *);
574 void (*sync_page)(struct page *);
576 /* Write back some dirty pages from this mapping. */
577 int (*writepages)(struct address_space *, struct writeback_control *);
579 /* Set a page dirty. Return true if this dirtied it */
580 int (*set_page_dirty)(struct page *page);
582 int (*readpages)(struct file *filp, struct address_space *mapping,
583 struct list_head *pages, unsigned nr_pages);
585 int (*write_begin)(struct file *, struct address_space *mapping,
586 loff_t pos, unsigned len, unsigned flags,
587 struct page **pagep, void **fsdata);
588 int (*write_end)(struct file *, struct address_space *mapping,
589 loff_t pos, unsigned len, unsigned copied,
590 struct page *page, void *fsdata);
592 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
593 sector_t (*bmap)(struct address_space *, sector_t);
594 void (*invalidatepage) (struct page *, unsigned long);
595 int (*releasepage) (struct page *, gfp_t);
596 ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
597 loff_t offset, unsigned long nr_segs);
598 int (*get_xip_mem)(struct address_space *, pgoff_t, int,
599 void **, unsigned long *);
600 /* migrate the contents of a page to the specified target */
601 int (*migratepage) (struct address_space *,
602 struct page *, struct page *);
603 int (*launder_page) (struct page *);
604 int (*is_partially_uptodate) (struct page *, read_descriptor_t *,
606 int (*error_remove_page)(struct address_space *, struct page *);
610 * pagecache_write_begin/pagecache_write_end must be used by general code
611 * to write into the pagecache.
613 int pagecache_write_begin(struct file *, struct address_space *mapping,
614 loff_t pos, unsigned len, unsigned flags,
615 struct page **pagep, void **fsdata);
617 int pagecache_write_end(struct file *, struct address_space *mapping,
618 loff_t pos, unsigned len, unsigned copied,
619 struct page *page, void *fsdata);
621 struct backing_dev_info;
622 struct address_space {
623 struct inode *host; /* owner: inode, block_device */
624 struct radix_tree_root page_tree; /* radix tree of all pages */
625 spinlock_t tree_lock; /* and lock protecting it */
626 unsigned int i_mmap_writable;/* count VM_SHARED mappings */
627 struct prio_tree_root i_mmap; /* tree of private and shared mappings */
628 struct list_head i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
629 spinlock_t i_mmap_lock; /* protect tree, count, list */
630 unsigned int truncate_count; /* Cover race condition with truncate */
631 unsigned long nrpages; /* number of total pages */
632 pgoff_t writeback_index;/* writeback starts here */
633 const struct address_space_operations *a_ops; /* methods */
634 unsigned long flags; /* error bits/gfp mask */
635 struct backing_dev_info *backing_dev_info; /* device readahead, etc */
636 spinlock_t private_lock; /* for use by the address_space */
637 struct list_head private_list; /* ditto */
638 struct address_space *assoc_mapping; /* ditto */
639 } __attribute__((aligned(sizeof(long))));
641 * On most architectures that alignment is already the case; but
642 * must be enforced here for CRIS, to let the least signficant bit
643 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
646 struct block_device {
647 dev_t bd_dev; /* not a kdev_t - it's a search key */
648 struct inode * bd_inode; /* will die */
649 struct super_block * bd_super;
651 struct mutex bd_mutex; /* open/close mutex */
652 struct list_head bd_inodes;
657 struct list_head bd_holder_list;
659 struct block_device * bd_contains;
660 unsigned bd_block_size;
661 struct hd_struct * bd_part;
662 /* number of times partitions within this device have been opened. */
663 unsigned bd_part_count;
665 struct gendisk * bd_disk;
666 struct list_head bd_list;
668 * Private data. You must have bd_claim'ed the block_device
669 * to use this. NOTE: bd_claim allows an owner to claim
670 * the same device multiple times, the owner must take special
671 * care to not mess up bd_private for that case.
673 unsigned long bd_private;
675 /* The counter of freeze processes */
676 int bd_fsfreeze_count;
677 /* Mutex for freeze */
678 struct mutex bd_fsfreeze_mutex;
682 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
685 #define PAGECACHE_TAG_DIRTY 0
686 #define PAGECACHE_TAG_WRITEBACK 1
687 #define PAGECACHE_TAG_TOWRITE 2
689 int mapping_tagged(struct address_space *mapping, int tag);
692 * Might pages of this file be mapped into userspace?
694 static inline int mapping_mapped(struct address_space *mapping)
696 return !prio_tree_empty(&mapping->i_mmap) ||
697 !list_empty(&mapping->i_mmap_nonlinear);
701 * Might pages of this file have been modified in userspace?
702 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
703 * marks vma as VM_SHARED if it is shared, and the file was opened for
704 * writing i.e. vma may be mprotected writable even if now readonly.
706 static inline int mapping_writably_mapped(struct address_space *mapping)
708 return mapping->i_mmap_writable != 0;
712 * Use sequence counter to get consistent i_size on 32-bit processors.
714 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
715 #include <linux/seqlock.h>
716 #define __NEED_I_SIZE_ORDERED
717 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
719 #define i_size_ordered_init(inode) do { } while (0)
723 #define ACL_NOT_CACHED ((void *)(-1))
726 struct hlist_node i_hash;
727 struct list_head i_list; /* backing dev IO list */
728 struct list_head i_sb_list;
729 struct list_head i_dentry;
732 unsigned int i_nlink;
736 unsigned int i_blkbits;
739 #ifdef __NEED_I_SIZE_ORDERED
740 seqcount_t i_size_seqcount;
742 struct timespec i_atime;
743 struct timespec i_mtime;
744 struct timespec i_ctime;
746 unsigned short i_bytes;
748 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
749 struct mutex i_mutex;
750 struct rw_semaphore i_alloc_sem;
751 const struct inode_operations *i_op;
752 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
753 struct super_block *i_sb;
754 struct file_lock *i_flock;
755 struct address_space *i_mapping;
756 struct address_space i_data;
758 struct dquot *i_dquot[MAXQUOTAS];
760 struct list_head i_devices;
762 struct pipe_inode_info *i_pipe;
763 struct block_device *i_bdev;
769 #ifdef CONFIG_FSNOTIFY
770 __u32 i_fsnotify_mask; /* all events this inode cares about */
771 struct hlist_head i_fsnotify_marks;
774 unsigned long i_state;
775 unsigned long dirtied_when; /* jiffies of first dirtying */
777 unsigned int i_flags;
779 atomic_t i_writecount;
780 #ifdef CONFIG_SECURITY
783 #ifdef CONFIG_FS_POSIX_ACL
784 struct posix_acl *i_acl;
785 struct posix_acl *i_default_acl;
787 void *i_private; /* fs or device private pointer */
791 * inode->i_mutex nesting subclasses for the lock validator:
793 * 0: the object of the current VFS operation
798 * The locking order between these classes is
799 * parent -> child -> normal -> xattr -> quota
801 enum inode_i_mutex_lock_class
811 * NOTE: in a 32bit arch with a preemptable kernel and
812 * an UP compile the i_size_read/write must be atomic
813 * with respect to the local cpu (unlike with preempt disabled),
814 * but they don't need to be atomic with respect to other cpus like in
815 * true SMP (so they need either to either locally disable irq around
816 * the read or for example on x86 they can be still implemented as a
817 * cmpxchg8b without the need of the lock prefix). For SMP compiles
818 * and 64bit archs it makes no difference if preempt is enabled or not.
820 static inline loff_t i_size_read(const struct inode *inode)
822 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
827 seq = read_seqcount_begin(&inode->i_size_seqcount);
828 i_size = inode->i_size;
829 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
831 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
835 i_size = inode->i_size;
839 return inode->i_size;
844 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
845 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
846 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
848 static inline void i_size_write(struct inode *inode, loff_t i_size)
850 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
851 write_seqcount_begin(&inode->i_size_seqcount);
852 inode->i_size = i_size;
853 write_seqcount_end(&inode->i_size_seqcount);
854 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
856 inode->i_size = i_size;
859 inode->i_size = i_size;
863 static inline unsigned iminor(const struct inode *inode)
865 return MINOR(inode->i_rdev);
868 static inline unsigned imajor(const struct inode *inode)
870 return MAJOR(inode->i_rdev);
873 extern struct block_device *I_BDEV(struct inode *inode);
876 rwlock_t lock; /* protects pid, uid, euid fields */
877 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
878 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
879 uid_t uid, euid; /* uid/euid of process setting the owner */
880 int signum; /* posix.1b rt signal to be delivered on IO */
884 * Track a single file's readahead state
886 struct file_ra_state {
887 pgoff_t start; /* where readahead started */
888 unsigned int size; /* # of readahead pages */
889 unsigned int async_size; /* do asynchronous readahead when
890 there are only # of pages ahead */
892 unsigned int ra_pages; /* Maximum readahead window */
893 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
894 loff_t prev_pos; /* Cache last read() position */
898 * Check if @index falls in the readahead windows.
900 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
902 return (index >= ra->start &&
903 index < ra->start + ra->size);
906 #define FILE_MNT_WRITE_TAKEN 1
907 #define FILE_MNT_WRITE_RELEASED 2
911 * fu_list becomes invalid after file_free is called and queued via
912 * fu_rcuhead for RCU freeing
915 struct list_head fu_list;
916 struct rcu_head fu_rcuhead;
919 #define f_dentry f_path.dentry
920 #define f_vfsmnt f_path.mnt
921 const struct file_operations *f_op;
922 spinlock_t f_lock; /* f_ep_links, f_flags, no IRQ */
926 atomic_long_t f_count;
927 unsigned int f_flags;
930 struct fown_struct f_owner;
931 const struct cred *f_cred;
932 struct file_ra_state f_ra;
935 #ifdef CONFIG_SECURITY
938 /* needed for tty driver, and maybe others */
942 /* Used by fs/eventpoll.c to link all the hooks to this file */
943 struct list_head f_ep_links;
944 #endif /* #ifdef CONFIG_EPOLL */
945 struct address_space *f_mapping;
946 #ifdef CONFIG_DEBUG_WRITECOUNT
947 unsigned long f_mnt_write_state;
951 #define get_file(x) atomic_long_inc(&(x)->f_count)
952 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
953 #define file_count(x) atomic_long_read(&(x)->f_count)
955 #ifdef CONFIG_DEBUG_WRITECOUNT
956 static inline void file_take_write(struct file *f)
958 WARN_ON(f->f_mnt_write_state != 0);
959 f->f_mnt_write_state = FILE_MNT_WRITE_TAKEN;
961 static inline void file_release_write(struct file *f)
963 f->f_mnt_write_state |= FILE_MNT_WRITE_RELEASED;
965 static inline void file_reset_write(struct file *f)
967 f->f_mnt_write_state = 0;
969 static inline void file_check_state(struct file *f)
972 * At this point, either both or neither of these bits
975 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN);
976 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_RELEASED);
978 static inline int file_check_writeable(struct file *f)
980 if (f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN)
982 printk(KERN_WARNING "writeable file with no "
983 "mnt_want_write()\n");
987 #else /* !CONFIG_DEBUG_WRITECOUNT */
988 static inline void file_take_write(struct file *filp) {}
989 static inline void file_release_write(struct file *filp) {}
990 static inline void file_reset_write(struct file *filp) {}
991 static inline void file_check_state(struct file *filp) {}
992 static inline int file_check_writeable(struct file *filp)
996 #endif /* CONFIG_DEBUG_WRITECOUNT */
998 #define MAX_NON_LFS ((1UL<<31) - 1)
1000 /* Page cache limit. The filesystems should put that into their s_maxbytes
1001 limits, otherwise bad things can happen in VM. */
1002 #if BITS_PER_LONG==32
1003 #define MAX_LFS_FILESIZE (((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
1004 #elif BITS_PER_LONG==64
1005 #define MAX_LFS_FILESIZE 0x7fffffffffffffffUL
1010 #define FL_ACCESS 8 /* not trying to lock, just looking */
1011 #define FL_EXISTS 16 /* when unlocking, test for existence */
1012 #define FL_LEASE 32 /* lease held on this file */
1013 #define FL_CLOSE 64 /* unlock on close */
1014 #define FL_SLEEP 128 /* A blocking lock */
1017 * Special return value from posix_lock_file() and vfs_lock_file() for
1018 * asynchronous locking.
1020 #define FILE_LOCK_DEFERRED 1
1023 * The POSIX file lock owner is determined by
1024 * the "struct files_struct" in the thread group
1025 * (or NULL for no owner - BSD locks).
1027 * Lockd stuffs a "host" pointer into this.
1029 typedef struct files_struct *fl_owner_t;
1031 struct file_lock_operations {
1032 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1033 void (*fl_release_private)(struct file_lock *);
1036 struct lock_manager_operations {
1037 int (*fl_compare_owner)(struct file_lock *, struct file_lock *);
1038 void (*fl_notify)(struct file_lock *); /* unblock callback */
1039 int (*fl_grant)(struct file_lock *, struct file_lock *, int);
1040 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1041 void (*fl_release_private)(struct file_lock *);
1042 void (*fl_break)(struct file_lock *);
1043 int (*fl_mylease)(struct file_lock *, struct file_lock *);
1044 int (*fl_change)(struct file_lock **, int);
1047 struct lock_manager {
1048 struct list_head list;
1051 void locks_start_grace(struct lock_manager *);
1052 void locks_end_grace(struct lock_manager *);
1053 int locks_in_grace(void);
1055 /* that will die - we need it for nfs_lock_info */
1056 #include <linux/nfs_fs_i.h>
1059 struct file_lock *fl_next; /* singly linked list for this inode */
1060 struct list_head fl_link; /* doubly linked list of all locks */
1061 struct list_head fl_block; /* circular list of blocked processes */
1062 fl_owner_t fl_owner;
1063 unsigned char fl_flags;
1064 unsigned char fl_type;
1065 unsigned int fl_pid;
1066 struct pid *fl_nspid;
1067 wait_queue_head_t fl_wait;
1068 struct file *fl_file;
1072 struct fasync_struct * fl_fasync; /* for lease break notifications */
1073 unsigned long fl_break_time; /* for nonblocking lease breaks */
1075 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1076 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1078 struct nfs_lock_info nfs_fl;
1079 struct nfs4_lock_info nfs4_fl;
1081 struct list_head link; /* link in AFS vnode's pending_locks list */
1082 int state; /* state of grant or error if -ve */
1087 /* The following constant reflects the upper bound of the file/locking space */
1089 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1090 #define OFFSET_MAX INT_LIMIT(loff_t)
1091 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1094 #include <linux/fcntl.h>
1096 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1098 #ifdef CONFIG_FILE_LOCKING
1099 extern int fcntl_getlk(struct file *, struct flock __user *);
1100 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1101 struct flock __user *);
1103 #if BITS_PER_LONG == 32
1104 extern int fcntl_getlk64(struct file *, struct flock64 __user *);
1105 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1106 struct flock64 __user *);
1109 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1110 extern int fcntl_getlease(struct file *filp);
1113 extern void locks_init_lock(struct file_lock *);
1114 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1115 extern void __locks_copy_lock(struct file_lock *, const struct file_lock *);
1116 extern void locks_remove_posix(struct file *, fl_owner_t);
1117 extern void locks_remove_flock(struct file *);
1118 extern void locks_release_private(struct file_lock *);
1119 extern void posix_test_lock(struct file *, struct file_lock *);
1120 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1121 extern int posix_lock_file_wait(struct file *, struct file_lock *);
1122 extern int posix_unblock_lock(struct file *, struct file_lock *);
1123 extern int vfs_test_lock(struct file *, struct file_lock *);
1124 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1125 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1126 extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
1127 extern int __break_lease(struct inode *inode, unsigned int flags);
1128 extern void lease_get_mtime(struct inode *, struct timespec *time);
1129 extern int generic_setlease(struct file *, long, struct file_lock **);
1130 extern int vfs_setlease(struct file *, long, struct file_lock **);
1131 extern int lease_modify(struct file_lock **, int);
1132 extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
1133 extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
1134 extern void lock_flocks(void);
1135 extern void unlock_flocks(void);
1136 #else /* !CONFIG_FILE_LOCKING */
1137 static inline int fcntl_getlk(struct file *file, struct flock __user *user)
1142 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1143 unsigned int cmd, struct flock __user *user)
1148 #if BITS_PER_LONG == 32
1149 static inline int fcntl_getlk64(struct file *file, struct flock64 __user *user)
1154 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1155 unsigned int cmd, struct flock64 __user *user)
1160 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1165 static inline int fcntl_getlease(struct file *filp)
1170 static inline void locks_init_lock(struct file_lock *fl)
1175 static inline void __locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1180 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1185 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1190 static inline void locks_remove_flock(struct file *filp)
1195 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1200 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1201 struct file_lock *conflock)
1206 static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1211 static inline int posix_unblock_lock(struct file *filp,
1212 struct file_lock *waiter)
1217 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1222 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1223 struct file_lock *fl, struct file_lock *conf)
1228 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1233 static inline int flock_lock_file_wait(struct file *filp,
1234 struct file_lock *request)
1239 static inline int __break_lease(struct inode *inode, unsigned int mode)
1244 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1249 static inline int generic_setlease(struct file *filp, long arg,
1250 struct file_lock **flp)
1255 static inline int vfs_setlease(struct file *filp, long arg,
1256 struct file_lock **lease)
1261 static inline int lease_modify(struct file_lock **before, int arg)
1266 static inline int lock_may_read(struct inode *inode, loff_t start,
1272 static inline int lock_may_write(struct inode *inode, loff_t start,
1278 static inline void lock_flocks(void)
1282 static inline void unlock_flocks(void)
1286 #endif /* !CONFIG_FILE_LOCKING */
1289 struct fasync_struct {
1293 struct fasync_struct *fa_next; /* singly linked list */
1294 struct file *fa_file;
1295 struct rcu_head fa_rcu;
1298 #define FASYNC_MAGIC 0x4601
1300 /* SMP safe fasync helpers: */
1301 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1302 /* can be called from interrupts */
1303 extern void kill_fasync(struct fasync_struct **, int, int);
1305 extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1306 extern int f_setown(struct file *filp, unsigned long arg, int force);
1307 extern void f_delown(struct file *filp);
1308 extern pid_t f_getown(struct file *filp);
1309 extern int send_sigurg(struct fown_struct *fown);
1315 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1316 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1317 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1318 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1319 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1321 extern struct list_head super_blocks;
1322 extern spinlock_t sb_lock;
1324 struct super_block {
1325 struct list_head s_list; /* Keep this first */
1326 dev_t s_dev; /* search index; _not_ kdev_t */
1327 unsigned char s_dirt;
1328 unsigned char s_blocksize_bits;
1329 unsigned long s_blocksize;
1330 loff_t s_maxbytes; /* Max file size */
1331 struct file_system_type *s_type;
1332 const struct super_operations *s_op;
1333 const struct dquot_operations *dq_op;
1334 const struct quotactl_ops *s_qcop;
1335 const struct export_operations *s_export_op;
1336 unsigned long s_flags;
1337 unsigned long s_magic;
1338 struct dentry *s_root;
1339 struct rw_semaphore s_umount;
1340 struct mutex s_lock;
1343 #ifdef CONFIG_SECURITY
1346 const struct xattr_handler **s_xattr;
1348 struct list_head s_inodes; /* all inodes */
1349 struct hlist_head s_anon; /* anonymous dentries for (nfs) exporting */
1351 struct list_head __percpu *s_files;
1353 struct list_head s_files;
1355 /* s_dentry_lru and s_nr_dentry_unused are protected by dcache_lock */
1356 struct list_head s_dentry_lru; /* unused dentry lru */
1357 int s_nr_dentry_unused; /* # of dentry on lru */
1359 struct block_device *s_bdev;
1360 struct backing_dev_info *s_bdi;
1361 struct mtd_info *s_mtd;
1362 struct list_head s_instances;
1363 struct quota_info s_dquot; /* Diskquota specific options */
1366 wait_queue_head_t s_wait_unfrozen;
1368 char s_id[32]; /* Informational name */
1370 void *s_fs_info; /* Filesystem private info */
1373 /* Granularity of c/m/atime in ns.
1374 Cannot be worse than a second */
1378 * The next field is for VFS *only*. No filesystems have any business
1379 * even looking at it. You had been warned.
1381 struct mutex s_vfs_rename_mutex; /* Kludge */
1384 * Filesystem subtype. If non-empty the filesystem type field
1385 * in /proc/mounts will be "type.subtype"
1390 * Saved mount options for lazy filesystems using
1391 * generic_show_options()
1393 char __rcu *s_options;
1396 extern struct timespec current_fs_time(struct super_block *sb);
1399 * Snapshotting support.
1403 SB_FREEZE_WRITE = 1,
1404 SB_FREEZE_TRANS = 2,
1407 #define vfs_check_frozen(sb, level) \
1408 wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level)))
1410 #define get_fs_excl() atomic_inc(¤t->fs_excl)
1411 #define put_fs_excl() atomic_dec(¤t->fs_excl)
1412 #define has_fs_excl() atomic_read(¤t->fs_excl)
1414 #define is_owner_or_cap(inode) \
1415 ((current_fsuid() == (inode)->i_uid) || capable(CAP_FOWNER))
1417 /* not quite ready to be deprecated, but... */
1418 extern void lock_super(struct super_block *);
1419 extern void unlock_super(struct super_block *);
1422 * VFS helper functions..
1424 extern int vfs_create(struct inode *, struct dentry *, int, struct nameidata *);
1425 extern int vfs_mkdir(struct inode *, struct dentry *, int);
1426 extern int vfs_mknod(struct inode *, struct dentry *, int, dev_t);
1427 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1428 extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
1429 extern int vfs_rmdir(struct inode *, struct dentry *);
1430 extern int vfs_unlink(struct inode *, struct dentry *);
1431 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1434 * VFS dentry helper functions.
1436 extern void dentry_unhash(struct dentry *dentry);
1439 * VFS file helper functions.
1441 extern int file_permission(struct file *, int);
1442 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1445 * VFS FS_IOC_FIEMAP helper definitions.
1447 struct fiemap_extent_info {
1448 unsigned int fi_flags; /* Flags as passed from user */
1449 unsigned int fi_extents_mapped; /* Number of mapped extents */
1450 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1451 struct fiemap_extent *fi_extents_start; /* Start of fiemap_extent
1454 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1455 u64 phys, u64 len, u32 flags);
1456 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1461 * NOTE! These match bits 12..15 of stat.st_mode
1462 * (ie "(i_mode >> 12) & 15").
1464 #define DT_UNKNOWN 0
1475 * This is the "filldir" function type, used by readdir() to let
1476 * the kernel specify what kind of dirent layout it wants to have.
1477 * This allows the kernel to read directories into kernel space or
1478 * to have different dirent layouts depending on the binary type.
1480 typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
1481 struct block_device_operations;
1483 /* These macros are for out of kernel modules to test that
1484 * the kernel supports the unlocked_ioctl and compat_ioctl
1485 * fields in struct file_operations. */
1486 #define HAVE_COMPAT_IOCTL 1
1487 #define HAVE_UNLOCKED_IOCTL 1
1491 * all file operations except setlease can be called without
1492 * the big kernel lock held in all filesystems.
1494 struct file_operations {
1495 struct module *owner;
1496 loff_t (*llseek) (struct file *, loff_t, int);
1497 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1498 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1499 ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1500 ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1501 int (*readdir) (struct file *, void *, filldir_t);
1502 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1503 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1504 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1505 int (*mmap) (struct file *, struct vm_area_struct *);
1506 int (*open) (struct inode *, struct file *);
1507 int (*flush) (struct file *, fl_owner_t id);
1508 int (*release) (struct inode *, struct file *);
1509 int (*fsync) (struct file *, int datasync);
1510 int (*aio_fsync) (struct kiocb *, int datasync);
1511 int (*fasync) (int, struct file *, int);
1512 int (*lock) (struct file *, int, struct file_lock *);
1513 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1514 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1515 int (*check_flags)(int);
1516 int (*flock) (struct file *, int, struct file_lock *);
1517 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1518 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1519 int (*setlease)(struct file *, long, struct file_lock **);
1522 struct inode_operations {
1523 int (*create) (struct inode *,struct dentry *,int, struct nameidata *);
1524 struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *);
1525 int (*link) (struct dentry *,struct inode *,struct dentry *);
1526 int (*unlink) (struct inode *,struct dentry *);
1527 int (*symlink) (struct inode *,struct dentry *,const char *);
1528 int (*mkdir) (struct inode *,struct dentry *,int);
1529 int (*rmdir) (struct inode *,struct dentry *);
1530 int (*mknod) (struct inode *,struct dentry *,int,dev_t);
1531 int (*rename) (struct inode *, struct dentry *,
1532 struct inode *, struct dentry *);
1533 int (*readlink) (struct dentry *, char __user *,int);
1534 void * (*follow_link) (struct dentry *, struct nameidata *);
1535 void (*put_link) (struct dentry *, struct nameidata *, void *);
1536 void (*truncate) (struct inode *);
1537 int (*permission) (struct inode *, int);
1538 int (*check_acl)(struct inode *, int);
1539 int (*setattr) (struct dentry *, struct iattr *);
1540 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1541 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1542 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1543 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1544 int (*removexattr) (struct dentry *, const char *);
1545 void (*truncate_range)(struct inode *, loff_t, loff_t);
1546 long (*fallocate)(struct inode *inode, int mode, loff_t offset,
1548 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1554 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1555 unsigned long nr_segs, unsigned long fast_segs,
1556 struct iovec *fast_pointer,
1557 struct iovec **ret_pointer);
1559 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1560 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1561 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1562 unsigned long, loff_t *);
1563 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1564 unsigned long, loff_t *);
1566 struct super_operations {
1567 struct inode *(*alloc_inode)(struct super_block *sb);
1568 void (*destroy_inode)(struct inode *);
1570 void (*dirty_inode) (struct inode *);
1571 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1572 int (*drop_inode) (struct inode *);
1573 void (*evict_inode) (struct inode *);
1574 void (*put_super) (struct super_block *);
1575 void (*write_super) (struct super_block *);
1576 int (*sync_fs)(struct super_block *sb, int wait);
1577 int (*freeze_fs) (struct super_block *);
1578 int (*unfreeze_fs) (struct super_block *);
1579 int (*statfs) (struct dentry *, struct kstatfs *);
1580 int (*remount_fs) (struct super_block *, int *, char *);
1581 void (*umount_begin) (struct super_block *);
1583 int (*show_options)(struct seq_file *, struct vfsmount *);
1584 int (*show_stats)(struct seq_file *, struct vfsmount *);
1586 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1587 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1589 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1593 * Inode state bits. Protected by inode_lock.
1595 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1596 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1598 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1599 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1600 * various stages of removing an inode.
1602 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1604 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1605 * fdatasync(). i_atime is the usual cause.
1606 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1607 * these changes separately from I_DIRTY_SYNC so that we
1608 * don't have to write inode on fdatasync() when only
1609 * mtime has changed in it.
1610 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1611 * I_NEW Serves as both a mutex and completion notification.
1612 * New inodes set I_NEW. If two processes both create
1613 * the same inode, one of them will release its inode and
1614 * wait for I_NEW to be released before returning.
1615 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1616 * also cause waiting on I_NEW, without I_NEW actually
1617 * being set. find_inode() uses this to prevent returning
1618 * nearly-dead inodes.
1619 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1620 * is zero. I_FREEING must be set when I_WILL_FREE is
1622 * I_FREEING Set when inode is about to be freed but still has dirty
1623 * pages or buffers attached or the inode itself is still
1625 * I_CLEAR Added by end_writeback(). In this state the inode is clean
1626 * and can be destroyed. Inode keeps I_FREEING.
1628 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1629 * prohibited for many purposes. iget() must wait for
1630 * the inode to be completely released, then create it
1631 * anew. Other functions will just ignore such inodes,
1632 * if appropriate. I_NEW is used for waiting.
1634 * I_SYNC Synchonized write of dirty inode data. The bits is
1635 * set during data writeback, and cleared with a wakeup
1636 * on the bit address once it is done.
1638 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1640 #define I_DIRTY_SYNC 1
1641 #define I_DIRTY_DATASYNC 2
1642 #define I_DIRTY_PAGES 4
1644 #define I_NEW (1 << __I_NEW)
1645 #define I_WILL_FREE 16
1646 #define I_FREEING 32
1649 #define I_SYNC (1 << __I_SYNC)
1651 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1653 extern void __mark_inode_dirty(struct inode *, int);
1654 static inline void mark_inode_dirty(struct inode *inode)
1656 __mark_inode_dirty(inode, I_DIRTY);
1659 static inline void mark_inode_dirty_sync(struct inode *inode)
1661 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1665 * inc_nlink - directly increment an inode's link count
1668 * This is a low-level filesystem helper to replace any
1669 * direct filesystem manipulation of i_nlink. Currently,
1670 * it is only here for parity with dec_nlink().
1672 static inline void inc_nlink(struct inode *inode)
1677 static inline void inode_inc_link_count(struct inode *inode)
1680 mark_inode_dirty(inode);
1684 * drop_nlink - directly drop an inode's link count
1687 * This is a low-level filesystem helper to replace any
1688 * direct filesystem manipulation of i_nlink. In cases
1689 * where we are attempting to track writes to the
1690 * filesystem, a decrement to zero means an imminent
1691 * write when the file is truncated and actually unlinked
1692 * on the filesystem.
1694 static inline void drop_nlink(struct inode *inode)
1700 * clear_nlink - directly zero an inode's link count
1703 * This is a low-level filesystem helper to replace any
1704 * direct filesystem manipulation of i_nlink. See
1705 * drop_nlink() for why we care about i_nlink hitting zero.
1707 static inline void clear_nlink(struct inode *inode)
1712 static inline void inode_dec_link_count(struct inode *inode)
1715 mark_inode_dirty(inode);
1719 * inode_inc_iversion - increments i_version
1720 * @inode: inode that need to be updated
1722 * Every time the inode is modified, the i_version field will be incremented.
1723 * The filesystem has to be mounted with i_version flag
1726 static inline void inode_inc_iversion(struct inode *inode)
1728 spin_lock(&inode->i_lock);
1730 spin_unlock(&inode->i_lock);
1733 extern void touch_atime(struct vfsmount *mnt, struct dentry *dentry);
1734 static inline void file_accessed(struct file *file)
1736 if (!(file->f_flags & O_NOATIME))
1737 touch_atime(file->f_path.mnt, file->f_path.dentry);
1740 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1742 struct file_system_type {
1745 int (*get_sb) (struct file_system_type *, int,
1746 const char *, void *, struct vfsmount *);
1747 void (*kill_sb) (struct super_block *);
1748 struct module *owner;
1749 struct file_system_type * next;
1750 struct list_head fs_supers;
1752 struct lock_class_key s_lock_key;
1753 struct lock_class_key s_umount_key;
1754 struct lock_class_key s_vfs_rename_key;
1756 struct lock_class_key i_lock_key;
1757 struct lock_class_key i_mutex_key;
1758 struct lock_class_key i_mutex_dir_key;
1759 struct lock_class_key i_alloc_sem_key;
1762 extern int get_sb_ns(struct file_system_type *fs_type, int flags, void *data,
1763 int (*fill_super)(struct super_block *, void *, int),
1764 struct vfsmount *mnt);
1765 extern int get_sb_bdev(struct file_system_type *fs_type,
1766 int flags, const char *dev_name, void *data,
1767 int (*fill_super)(struct super_block *, void *, int),
1768 struct vfsmount *mnt);
1769 extern int get_sb_single(struct file_system_type *fs_type,
1770 int flags, void *data,
1771 int (*fill_super)(struct super_block *, void *, int),
1772 struct vfsmount *mnt);
1773 extern int get_sb_nodev(struct file_system_type *fs_type,
1774 int flags, void *data,
1775 int (*fill_super)(struct super_block *, void *, int),
1776 struct vfsmount *mnt);
1777 void generic_shutdown_super(struct super_block *sb);
1778 void kill_block_super(struct super_block *sb);
1779 void kill_anon_super(struct super_block *sb);
1780 void kill_litter_super(struct super_block *sb);
1781 void deactivate_super(struct super_block *sb);
1782 void deactivate_locked_super(struct super_block *sb);
1783 int set_anon_super(struct super_block *s, void *data);
1784 struct super_block *sget(struct file_system_type *type,
1785 int (*test)(struct super_block *,void *),
1786 int (*set)(struct super_block *,void *),
1788 extern int get_sb_pseudo(struct file_system_type *, char *,
1789 const struct super_operations *ops, unsigned long,
1790 struct vfsmount *mnt);
1791 extern void simple_set_mnt(struct vfsmount *mnt, struct super_block *sb);
1793 static inline void sb_mark_dirty(struct super_block *sb)
1797 static inline void sb_mark_clean(struct super_block *sb)
1801 static inline int sb_is_dirty(struct super_block *sb)
1806 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1807 #define fops_get(fops) \
1808 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1809 #define fops_put(fops) \
1810 do { if (fops) module_put((fops)->owner); } while(0)
1812 extern int register_filesystem(struct file_system_type *);
1813 extern int unregister_filesystem(struct file_system_type *);
1814 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1815 #define kern_mount(type) kern_mount_data(type, NULL)
1816 extern int may_umount_tree(struct vfsmount *);
1817 extern int may_umount(struct vfsmount *);
1818 extern long do_mount(char *, char *, char *, unsigned long, void *);
1819 extern struct vfsmount *collect_mounts(struct path *);
1820 extern void drop_collected_mounts(struct vfsmount *);
1821 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1823 extern int vfs_statfs(struct path *, struct kstatfs *);
1824 extern int statfs_by_dentry(struct dentry *, struct kstatfs *);
1825 extern int freeze_super(struct super_block *super);
1826 extern int thaw_super(struct super_block *super);
1828 extern int current_umask(void);
1831 extern struct kobject *fs_kobj;
1833 extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1835 #define FLOCK_VERIFY_READ 1
1836 #define FLOCK_VERIFY_WRITE 2
1838 #ifdef CONFIG_FILE_LOCKING
1839 extern int locks_mandatory_locked(struct inode *);
1840 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1843 * Candidates for mandatory locking have the setgid bit set
1844 * but no group execute bit - an otherwise meaningless combination.
1847 static inline int __mandatory_lock(struct inode *ino)
1849 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
1853 * ... and these candidates should be on MS_MANDLOCK mounted fs,
1854 * otherwise these will be advisory locks
1857 static inline int mandatory_lock(struct inode *ino)
1859 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
1862 static inline int locks_verify_locked(struct inode *inode)
1864 if (mandatory_lock(inode))
1865 return locks_mandatory_locked(inode);
1869 static inline int locks_verify_truncate(struct inode *inode,
1873 if (inode->i_flock && mandatory_lock(inode))
1874 return locks_mandatory_area(
1875 FLOCK_VERIFY_WRITE, inode, filp,
1876 size < inode->i_size ? size : inode->i_size,
1877 (size < inode->i_size ? inode->i_size - size
1878 : size - inode->i_size)
1883 static inline int break_lease(struct inode *inode, unsigned int mode)
1886 return __break_lease(inode, mode);
1889 #else /* !CONFIG_FILE_LOCKING */
1890 static inline int locks_mandatory_locked(struct inode *inode)
1895 static inline int locks_mandatory_area(int rw, struct inode *inode,
1896 struct file *filp, loff_t offset,
1902 static inline int __mandatory_lock(struct inode *inode)
1907 static inline int mandatory_lock(struct inode *inode)
1912 static inline int locks_verify_locked(struct inode *inode)
1917 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
1923 static inline int break_lease(struct inode *inode, unsigned int mode)
1928 #endif /* CONFIG_FILE_LOCKING */
1932 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
1934 extern int do_fallocate(struct file *file, int mode, loff_t offset,
1936 extern long do_sys_open(int dfd, const char __user *filename, int flags,
1938 extern struct file *filp_open(const char *, int, int);
1939 extern struct file * dentry_open(struct dentry *, struct vfsmount *, int,
1940 const struct cred *);
1941 extern int filp_close(struct file *, fl_owner_t id);
1942 extern char * getname(const char __user *);
1946 extern int ioctl_preallocate(struct file *filp, void __user *argp);
1949 extern void __init vfs_caches_init_early(void);
1950 extern void __init vfs_caches_init(unsigned long);
1952 extern struct kmem_cache *names_cachep;
1954 #define __getname_gfp(gfp) kmem_cache_alloc(names_cachep, (gfp))
1955 #define __getname() __getname_gfp(GFP_KERNEL)
1956 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
1957 #ifndef CONFIG_AUDITSYSCALL
1958 #define putname(name) __putname(name)
1960 extern void putname(const char *name);
1964 extern int register_blkdev(unsigned int, const char *);
1965 extern void unregister_blkdev(unsigned int, const char *);
1966 extern struct block_device *bdget(dev_t);
1967 extern struct block_device *bdgrab(struct block_device *bdev);
1968 extern void bd_set_size(struct block_device *, loff_t size);
1969 extern void bd_forget(struct inode *inode);
1970 extern void bdput(struct block_device *);
1971 extern struct block_device *open_by_devnum(dev_t, fmode_t);
1972 extern void invalidate_bdev(struct block_device *);
1973 extern int sync_blockdev(struct block_device *bdev);
1974 extern struct super_block *freeze_bdev(struct block_device *);
1975 extern void emergency_thaw_all(void);
1976 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
1977 extern int fsync_bdev(struct block_device *);
1979 static inline void bd_forget(struct inode *inode) {}
1980 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
1981 static inline void invalidate_bdev(struct block_device *bdev) {}
1983 static inline struct super_block *freeze_bdev(struct block_device *sb)
1988 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
1993 extern int sync_filesystem(struct super_block *);
1994 extern const struct file_operations def_blk_fops;
1995 extern const struct file_operations def_chr_fops;
1996 extern const struct file_operations bad_sock_fops;
1997 extern const struct file_operations def_fifo_fops;
1999 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2000 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2001 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2002 extern int blkdev_get(struct block_device *, fmode_t);
2003 extern int blkdev_put(struct block_device *, fmode_t);
2004 extern int bd_claim(struct block_device *, void *);
2005 extern void bd_release(struct block_device *);
2007 extern int bd_claim_by_disk(struct block_device *, void *, struct gendisk *);
2008 extern void bd_release_from_disk(struct block_device *, struct gendisk *);
2010 #define bd_claim_by_disk(bdev, holder, disk) bd_claim(bdev, holder)
2011 #define bd_release_from_disk(bdev, disk) bd_release(bdev)
2016 #define CHRDEV_MAJOR_HASH_SIZE 255
2017 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2018 extern int register_chrdev_region(dev_t, unsigned, const char *);
2019 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2020 unsigned int count, const char *name,
2021 const struct file_operations *fops);
2022 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2023 unsigned int count, const char *name);
2024 extern void unregister_chrdev_region(dev_t, unsigned);
2025 extern void chrdev_show(struct seq_file *,off_t);
2027 static inline int register_chrdev(unsigned int major, const char *name,
2028 const struct file_operations *fops)
2030 return __register_chrdev(major, 0, 256, name, fops);
2033 static inline void unregister_chrdev(unsigned int major, const char *name)
2035 __unregister_chrdev(major, 0, 256, name);
2038 /* fs/block_dev.c */
2039 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2040 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2043 #define BLKDEV_MAJOR_HASH_SIZE 255
2044 extern const char *__bdevname(dev_t, char *buffer);
2045 extern const char *bdevname(struct block_device *bdev, char *buffer);
2046 extern struct block_device *lookup_bdev(const char *);
2047 extern struct block_device *open_bdev_exclusive(const char *, fmode_t, void *);
2048 extern void close_bdev_exclusive(struct block_device *, fmode_t);
2049 extern void blkdev_show(struct seq_file *,off_t);
2052 #define BLKDEV_MAJOR_HASH_SIZE 0
2055 extern void init_special_inode(struct inode *, umode_t, dev_t);
2057 /* Invalid inode operations -- fs/bad_inode.c */
2058 extern void make_bad_inode(struct inode *);
2059 extern int is_bad_inode(struct inode *);
2061 extern const struct file_operations read_pipefifo_fops;
2062 extern const struct file_operations write_pipefifo_fops;
2063 extern const struct file_operations rdwr_pipefifo_fops;
2065 extern int fs_may_remount_ro(struct super_block *);
2069 * return READ, READA, or WRITE
2071 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2074 * return data direction, READ or WRITE
2076 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2078 extern void check_disk_size_change(struct gendisk *disk,
2079 struct block_device *bdev);
2080 extern int revalidate_disk(struct gendisk *);
2081 extern int check_disk_change(struct block_device *);
2082 extern int __invalidate_device(struct block_device *);
2083 extern int invalidate_partition(struct gendisk *, int);
2085 extern int invalidate_inodes(struct super_block *);
2086 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2087 pgoff_t start, pgoff_t end);
2089 static inline void invalidate_remote_inode(struct inode *inode)
2091 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2092 S_ISLNK(inode->i_mode))
2093 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2095 extern int invalidate_inode_pages2(struct address_space *mapping);
2096 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2097 pgoff_t start, pgoff_t end);
2098 extern int write_inode_now(struct inode *, int);
2099 extern int filemap_fdatawrite(struct address_space *);
2100 extern int filemap_flush(struct address_space *);
2101 extern int filemap_fdatawait(struct address_space *);
2102 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2104 extern int filemap_write_and_wait(struct address_space *mapping);
2105 extern int filemap_write_and_wait_range(struct address_space *mapping,
2106 loff_t lstart, loff_t lend);
2107 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2108 loff_t start, loff_t end, int sync_mode);
2109 extern int filemap_fdatawrite_range(struct address_space *mapping,
2110 loff_t start, loff_t end);
2112 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2114 extern int vfs_fsync(struct file *file, int datasync);
2115 extern int generic_write_sync(struct file *file, loff_t pos, loff_t count);
2116 extern void sync_supers(void);
2117 extern void emergency_sync(void);
2118 extern void emergency_remount(void);
2120 extern sector_t bmap(struct inode *, sector_t);
2122 extern int notify_change(struct dentry *, struct iattr *);
2123 extern int inode_permission(struct inode *, int);
2124 extern int generic_permission(struct inode *, int,
2125 int (*check_acl)(struct inode *, int));
2127 static inline bool execute_ok(struct inode *inode)
2129 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2132 extern int get_write_access(struct inode *);
2133 extern int deny_write_access(struct file *);
2134 static inline void put_write_access(struct inode * inode)
2136 atomic_dec(&inode->i_writecount);
2138 static inline void allow_write_access(struct file *file)
2141 atomic_inc(&file->f_path.dentry->d_inode->i_writecount);
2143 extern int do_pipe_flags(int *, int);
2144 extern struct file *create_read_pipe(struct file *f, int flags);
2145 extern struct file *create_write_pipe(int flags);
2146 extern void free_write_pipe(struct file *);
2148 extern struct file *do_filp_open(int dfd, const char *pathname,
2149 int open_flag, int mode, int acc_mode);
2150 extern int may_open(struct path *, int, int);
2152 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2153 extern struct file * open_exec(const char *);
2155 /* fs/dcache.c -- generic fs support functions */
2156 extern int is_subdir(struct dentry *, struct dentry *);
2157 extern int path_is_under(struct path *, struct path *);
2158 extern ino_t find_inode_number(struct dentry *, struct qstr *);
2160 #include <linux/err.h>
2162 /* needed for stackable file system support */
2163 extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
2165 extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
2167 extern int inode_init_always(struct super_block *, struct inode *);
2168 extern void inode_init_once(struct inode *);
2169 extern void inode_add_to_lists(struct super_block *, struct inode *);
2170 extern void iput(struct inode *);
2171 extern struct inode * igrab(struct inode *);
2172 extern ino_t iunique(struct super_block *, ino_t);
2173 extern int inode_needs_sync(struct inode *inode);
2174 extern int generic_delete_inode(struct inode *inode);
2175 extern int generic_drop_inode(struct inode *inode);
2177 extern struct inode *ilookup5_nowait(struct super_block *sb,
2178 unsigned long hashval, int (*test)(struct inode *, void *),
2180 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2181 int (*test)(struct inode *, void *), void *data);
2182 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2184 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2185 extern struct inode * iget_locked(struct super_block *, unsigned long);
2186 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2187 extern int insert_inode_locked(struct inode *);
2188 extern void unlock_new_inode(struct inode *);
2190 extern void __iget(struct inode * inode);
2191 extern void iget_failed(struct inode *);
2192 extern void end_writeback(struct inode *);
2193 extern void destroy_inode(struct inode *);
2194 extern void __destroy_inode(struct inode *);
2195 extern struct inode *new_inode(struct super_block *);
2196 extern int should_remove_suid(struct dentry *);
2197 extern int file_remove_suid(struct file *);
2199 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2200 extern void remove_inode_hash(struct inode *);
2201 static inline void insert_inode_hash(struct inode *inode) {
2202 __insert_inode_hash(inode, inode->i_ino);
2206 extern void submit_bio(int, struct bio *);
2207 extern int bdev_read_only(struct block_device *);
2209 extern int set_blocksize(struct block_device *, int);
2210 extern int sb_set_blocksize(struct super_block *, int);
2211 extern int sb_min_blocksize(struct super_block *, int);
2213 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2214 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2215 extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
2216 int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
2217 extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2218 extern ssize_t __generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long,
2220 extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2221 extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
2222 unsigned long *, loff_t, loff_t *, size_t, size_t);
2223 extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
2224 unsigned long, loff_t, loff_t *, size_t, ssize_t);
2225 extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2226 extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
2227 extern int generic_segment_checks(const struct iovec *iov,
2228 unsigned long *nr_segs, size_t *count, int access_flags);
2230 /* fs/block_dev.c */
2231 extern ssize_t blkdev_aio_write(struct kiocb *iocb, const struct iovec *iov,
2232 unsigned long nr_segs, loff_t pos);
2233 extern int blkdev_fsync(struct file *filp, int datasync);
2236 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2237 struct pipe_inode_info *, size_t, unsigned int);
2238 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2239 struct pipe_inode_info *, size_t, unsigned int);
2240 extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
2241 struct file *, loff_t *, size_t, unsigned int);
2242 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2243 struct file *out, loff_t *, size_t len, unsigned int flags);
2244 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2245 size_t len, unsigned int flags);
2248 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2249 extern loff_t noop_llseek(struct file *file, loff_t offset, int origin);
2250 extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
2251 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
2252 extern loff_t generic_file_llseek_unlocked(struct file *file, loff_t offset,
2254 extern int generic_file_open(struct inode * inode, struct file * filp);
2255 extern int nonseekable_open(struct inode * inode, struct file * filp);
2257 #ifdef CONFIG_FS_XIP
2258 extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
2260 extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
2261 extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
2262 size_t len, loff_t *ppos);
2263 extern int xip_truncate_page(struct address_space *mapping, loff_t from);
2265 static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
2272 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2273 loff_t file_offset);
2276 /* need locking between buffered and direct access */
2279 /* filesystem does not support filling holes */
2280 DIO_SKIP_HOLES = 0x02,
2283 void dio_end_io(struct bio *bio, int error);
2285 ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
2286 struct block_device *bdev, const struct iovec *iov, loff_t offset,
2287 unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
2288 dio_submit_t submit_io, int flags);
2290 static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
2291 struct inode *inode, struct block_device *bdev, const struct iovec *iov,
2292 loff_t offset, unsigned long nr_segs, get_block_t get_block,
2293 dio_iodone_t end_io)
2295 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
2296 nr_segs, get_block, end_io, NULL,
2297 DIO_LOCKING | DIO_SKIP_HOLES);
2301 extern const struct file_operations generic_ro_fops;
2303 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2305 extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
2306 extern int vfs_follow_link(struct nameidata *, const char *);
2307 extern int page_readlink(struct dentry *, char __user *, int);
2308 extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2309 extern void page_put_link(struct dentry *, struct nameidata *, void *);
2310 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2312 extern int page_symlink(struct inode *inode, const char *symname, int len);
2313 extern const struct inode_operations page_symlink_inode_operations;
2314 extern int generic_readlink(struct dentry *, char __user *, int);
2315 extern void generic_fillattr(struct inode *, struct kstat *);
2316 extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2317 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2318 void inode_add_bytes(struct inode *inode, loff_t bytes);
2319 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2320 loff_t inode_get_bytes(struct inode *inode);
2321 void inode_set_bytes(struct inode *inode, loff_t bytes);
2323 extern int vfs_readdir(struct file *, filldir_t, void *);
2325 extern int vfs_stat(const char __user *, struct kstat *);
2326 extern int vfs_lstat(const char __user *, struct kstat *);
2327 extern int vfs_fstat(unsigned int, struct kstat *);
2328 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2330 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2332 extern int __generic_block_fiemap(struct inode *inode,
2333 struct fiemap_extent_info *fieinfo,
2334 loff_t start, loff_t len,
2335 get_block_t *get_block);
2336 extern int generic_block_fiemap(struct inode *inode,
2337 struct fiemap_extent_info *fieinfo, u64 start,
2338 u64 len, get_block_t *get_block);
2340 extern void get_filesystem(struct file_system_type *fs);
2341 extern void put_filesystem(struct file_system_type *fs);
2342 extern struct file_system_type *get_fs_type(const char *name);
2343 extern struct super_block *get_super(struct block_device *);
2344 extern struct super_block *get_active_super(struct block_device *bdev);
2345 extern struct super_block *user_get_super(dev_t);
2346 extern void drop_super(struct super_block *sb);
2347 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2349 extern int dcache_dir_open(struct inode *, struct file *);
2350 extern int dcache_dir_close(struct inode *, struct file *);
2351 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2352 extern int dcache_readdir(struct file *, void *, filldir_t);
2353 extern int simple_setattr(struct dentry *, struct iattr *);
2354 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2355 extern int simple_statfs(struct dentry *, struct kstatfs *);
2356 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2357 extern int simple_unlink(struct inode *, struct dentry *);
2358 extern int simple_rmdir(struct inode *, struct dentry *);
2359 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2360 extern int noop_fsync(struct file *, int);
2361 extern int simple_empty(struct dentry *);
2362 extern int simple_readpage(struct file *file, struct page *page);
2363 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2364 loff_t pos, unsigned len, unsigned flags,
2365 struct page **pagep, void **fsdata);
2366 extern int simple_write_end(struct file *file, struct address_space *mapping,
2367 loff_t pos, unsigned len, unsigned copied,
2368 struct page *page, void *fsdata);
2370 extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *);
2371 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2372 extern const struct file_operations simple_dir_operations;
2373 extern const struct inode_operations simple_dir_inode_operations;
2374 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2375 struct dentry *d_alloc_name(struct dentry *, const char *);
2376 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2377 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2378 extern void simple_release_fs(struct vfsmount **mount, int *count);
2380 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2381 loff_t *ppos, const void *from, size_t available);
2382 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2383 const void __user *from, size_t count);
2385 extern int generic_file_fsync(struct file *, int);
2387 extern int generic_check_addressable(unsigned, u64);
2389 #ifdef CONFIG_MIGRATION
2390 extern int buffer_migrate_page(struct address_space *,
2391 struct page *, struct page *);
2393 #define buffer_migrate_page NULL
2396 extern int inode_change_ok(const struct inode *, struct iattr *);
2397 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2398 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2400 extern void file_update_time(struct file *file);
2402 extern int generic_show_options(struct seq_file *m, struct vfsmount *mnt);
2403 extern void save_mount_options(struct super_block *sb, char *options);
2404 extern void replace_mount_options(struct super_block *sb, char *options);
2406 static inline ino_t parent_ino(struct dentry *dentry)
2410 spin_lock(&dentry->d_lock);
2411 res = dentry->d_parent->d_inode->i_ino;
2412 spin_unlock(&dentry->d_lock);
2416 /* Transaction based IO helpers */
2419 * An argresp is stored in an allocated page and holds the
2420 * size of the argument or response, along with its content
2422 struct simple_transaction_argresp {
2427 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2429 char *simple_transaction_get(struct file *file, const char __user *buf,
2431 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2432 size_t size, loff_t *pos);
2433 int simple_transaction_release(struct inode *inode, struct file *file);
2435 void simple_transaction_set(struct file *file, size_t n);
2438 * simple attribute files
2440 * These attributes behave similar to those in sysfs:
2442 * Writing to an attribute immediately sets a value, an open file can be
2443 * written to multiple times.
2445 * Reading from an attribute creates a buffer from the value that might get
2446 * read with multiple read calls. When the attribute has been read
2447 * completely, no further read calls are possible until the file is opened
2450 * All attributes contain a text representation of a numeric value
2451 * that are accessed with the get() and set() functions.
2453 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2454 static int __fops ## _open(struct inode *inode, struct file *file) \
2456 __simple_attr_check_format(__fmt, 0ull); \
2457 return simple_attr_open(inode, file, __get, __set, __fmt); \
2459 static const struct file_operations __fops = { \
2460 .owner = THIS_MODULE, \
2461 .open = __fops ## _open, \
2462 .release = simple_attr_release, \
2463 .read = simple_attr_read, \
2464 .write = simple_attr_write, \
2465 .llseek = generic_file_llseek, \
2468 static inline void __attribute__((format(printf, 1, 2)))
2469 __simple_attr_check_format(const char *fmt, ...)
2471 /* don't do anything, just let the compiler check the arguments; */
2474 int simple_attr_open(struct inode *inode, struct file *file,
2475 int (*get)(void *, u64 *), int (*set)(void *, u64),
2477 int simple_attr_release(struct inode *inode, struct file *file);
2478 ssize_t simple_attr_read(struct file *file, char __user *buf,
2479 size_t len, loff_t *ppos);
2480 ssize_t simple_attr_write(struct file *file, const char __user *buf,
2481 size_t len, loff_t *ppos);
2484 int proc_nr_files(struct ctl_table *table, int write,
2485 void __user *buffer, size_t *lenp, loff_t *ppos);
2487 int __init get_filesystem_list(char *buf);
2489 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2490 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2491 (flag & FMODE_NONOTIFY)))
2493 #endif /* __KERNEL__ */
2494 #endif /* _LINUX_FS_H */